US9674521B2ActiveUtilityA1

Prediction method and apparatus for chroma component of image using luma component of image

96
Assignee: LEE TAMMYPriority: Jun 28, 2011Filed: Jun 28, 2012Granted: Jun 6, 2017
Est. expiryJun 28, 2031(~5 yrs left)· nominal 20-yr term from priority
H04N 19/593H04N 19/186H04N 19/50H04N 19/117H04N 9/77H04N 19/59H04N 19/159H04N 19/182H04N 19/70H04N 19/00315
96
PatentIndex Score
34
Cited by
46
References
8
Claims

Abstract

Provided are a method and an apparatus for predicting a chrominance signal from a luminance signal by using a correlation between the luminance and chrominance signals. A method of predicting a chrominance component image generates a down-sampled luminance signal matching a chrominance signal by independently applying various down-sampling methods according to a location of a luminance signal, and predicts a chrominance signal from the down-sampled luminance signal, while considering a size of a luminance prediction unit and a size of a chrominance prediction unit corresponding to each other according to a color format.

Claims

exact text as granted — not AI-modified
What is claimed is: 
     
       1. A method of predicting a chrominance component image by using a luminance component image, the method comprising:
 down-sampling inner pixels of a luminance prediction unit and adjacent pixels of the luminance prediction unit, the luminance prediction unit having been pre-encoded and pre-restored, considering a size of a luminance prediction unit and a size of a chrominance prediction unit corresponding to each other according to a color format; 
 obtaining parameters indicating a correlation between the luminance prediction unit and a predicted chrominance prediction unit based on the down-sampled adjacent pixels of the luminance prediction unit and restored adjacent pixels of the predicted chrominance prediction unit; and 
 obtaining, from the down-sampled luminance prediction unit by using the obtained parameters, a prediction value of a chrominance prediction unit corresponding to the down-sampled luminance prediction unit, 
 wherein the down-sampling comprises:
 selecting a first location of a luminance pixel to be selected from a predetermined luminance pixel group or a first filtering method of luminance pixels, for down-sampling with respect to the inner pixels of the luminance prediction unit; 
 selecting a second location of a luminance pixel to be selected from the predetermined luminance pixel group or a second filtering method of luminance pixels, for down-sampling with respect to adjacent upper pixels of the luminance prediction unit; and 
 selecting a third location of a luminance pixel to be selected from the predetermined luminance pixel group or a third filtering method of luminance pixels, for down-sampling with respect to adjacent left pixels of the luminance prediction unit, 
 
 wherein the first location and first filtering method used for down-sampling with respect to the inner pixels of the luminance prediction unit, the second location and second filtering method used for down-sampling with respect to the adjacent upper pixels of the luminance prediction unit, and the third location and third filtering method used for down-sampling with respect to the adjacent left pixels of the luminance prediction unit are independently selected, 
 wherein, when the size of the luminance prediction unit is 2nS×2nS where nS is a positive integer:
 the chrominance prediction unit corresponding to the luminance prediction unit is nS×nS, an inner pixel of the luminance prediction unit is RecL(x,y) where x and y are each from 0 to 2nS-1, an upper pixel of the luminance prediction unit is RecL(x,−1) where x is from 0 to 2nS-1, a left pixel of the luminance prediction unit is RecL(−1,y) where y is from 0 to 2nS-1, and a down-sampled luminance pixel corresponding to a chrominance pixel at (x,y) is RecL′(x,y); 
 a down-sampled luminance pixel RecL(x,−1) corresponding to an adjacent upper pixel Pc(x,−1) of the chrominance prediction unit is obtained according to equation, RecL′(x,−1)=(RecL(2x−1,−1)+2*RecL(2x,−1)+RecL(2x+1,−1)+2)>>2 or RecL′(x,−1)=RecL(2x, −1), with respect to adjacent upper pixels of the luminance prediction unit; 
 a down-sampled luminance pixel RecL′(−1,y) corresponding to an adjacent left pixel Pc(−1,y) of the chrominance prediction unit is obtained according to equation, RecL′(−1,y)=(RecL(−1,2y)+RecL(−1,2y=1))>>1, with respect to adjacent left pixels of the luminance prediction unit; and 
 a down-sampled luminance pixel RecL′(x,y) corresponding to inner pixels Pc(x,y) of the chrominance prediction unit is obtained according to equation, RecL′(x,y)=(RecL(2x,2y)+RecL(2x,2y+1))>>1, with respect to the inner pixels of the luminance prediction unit. 
 
 
     
     
       2. The method of  claim 1 , wherein the parameters comprise a weight and offset value defining a linear relationship between a restored pixel of the luminance prediction unit and a corresponding pixel of the chrominance prediction unit. 
     
     
       3. The method of  claim 2 , wherein, when a restored pixel at a down-sampled (x,y) location of the luminance prediction unit is Rec 13 L′(x,y):
 a prediction value of a pixel at a (x,y) location of the chrominance prediction unit is Pred 13 C(x,y), the weight is a, and the offset value is b; and 
 the correlation is defined according to equation, Pred 13 C(x,y)=a*Rec L′(x,y)+b. 
 
     
     
       4. The method of  claim 3 , wherein, when I adjacent restored pixels of the luminance prediction unit are Rec_L′(i) where I is an integer from 0 to I-1, and I adjacent restored pixels of the chrominance prediction unit are Rec_C(i),
 the weight a is determined according to equation, 
 
       
         
           
             
               
                 
                   
                     a 
                     = 
                       
                     ⁢ 
                     
                       
                         
                           I 
                           · 
                           
                             
                               ∑ 
                               
                                 i 
                                 = 
                                 0 
                               
                               
                                 I 
                                 ⁢ 
                                 
                                   - 
                                 
                                 ⁢ 
                                 1 
                               
                             
                             ⁢ 
                             
                               Rec_c 
                               ⁢ 
                               
                                 
                                   ( 
                                   i 
                                   ) 
                                 
                                 · 
                                 
                                   Rec_L 
                                   ′ 
                                 
                               
                               ⁢ 
                               
                                 ( 
                                 i 
                                 ) 
                               
                             
                           
                         
                         - 
                         
                           
                             ∑ 
                             
                               i 
                               = 
                               0 
                             
                             
                               I 
                               ⁢ 
                               
                                 - 
                               
                               ⁢ 
                               1 
                             
                           
                           ⁢ 
                           
                             Rec_c 
                             ⁢ 
                             
                               
                                 ( 
                                 i 
                                 ) 
                               
                               · 
                               
                                 
                                   ∑ 
                                   
                                     i 
                                     = 
                                     0 
                                   
                                   
                                     I 
                                     ⁢ 
                                     
                                       - 
                                     
                                     ⁢ 
                                     1 
                                   
                                 
                                 ⁢ 
                                 
                                   
                                     Rec_L 
                                     ′ 
                                   
                                   ⁢ 
                                   
                                     ( 
                                     i 
                                     ) 
                                   
                                 
                               
                             
                           
                         
                       
                       
                         
                           I 
                           · 
                           
                             
                               ∑ 
                               
                                 i 
                                 = 
                                 0 
                               
                               
                                 I 
                                 ⁢ 
                                 
                                   - 
                                 
                                 ⁢ 
                                 1 
                               
                             
                             ⁢ 
                             
                               
                                 Rec_L 
                                 ′ 
                               
                               ⁢ 
                               
                                 
                                   ( 
                                   i 
                                   ) 
                                 
                                 · 
                                 
                                   Rec_L 
                                   ′ 
                                 
                               
                               ⁢ 
                               
                                 ( 
                                 i 
                                 ) 
                               
                             
                           
                         
                         - 
                         
                           
                             ( 
                             
                               
                                 ∑ 
                                 
                                   i 
                                   = 
                                   0 
                                 
                                 
                                   I 
                                   ⁢ 
                                   
                                     - 
                                   
                                   ⁢ 
                                   1 
                                 
                               
                               ⁢ 
                               
                                 
                                   Rec_L 
                                   ′ 
                                 
                                 ⁢ 
                                 
                                   ( 
                                   i 
                                   ) 
                                 
                               
                             
                             ) 
                           
                           2 
                         
                       
                     
                   
                 
               
               
                 
                   
                     = 
                       
                     ⁢ 
                     
                       
                         
                           A 
                           ⁢ 
                           
                               
                           
                           ⁢ 
                           1 
                         
                         
                           A 
                           ⁢ 
                           
                               
                           
                           ⁢ 
                           2 
                         
                       
                       . 
                     
                   
                 
               
             
           
         
       
     
     
       5. The method of  claim 3 , wherein the offset value b is determined according to equation, 
       
         
           
             
               b 
               = 
               
                 
                   
                     
                       
                         ∑ 
                         
                           i 
                           = 
                           0 
                         
                         
                           I 
                           ⁢ 
                           
                             - 
                           
                           ⁢ 
                           1 
                         
                       
                       ⁢ 
                       
                         Rec_c 
                         ⁢ 
                         
                           ( 
                           i 
                           ) 
                         
                       
                     
                     - 
                     
                       a 
                       · 
                       
                         
                           ∑ 
                           
                             i 
                             = 
                             0 
                           
                           
                             I 
                             ⁢ 
                             
                               - 
                             
                             ⁢ 
                             1 
                           
                         
                         ⁢ 
                         
                           
                             Rec_L 
                             
                               ′ 
                               ⁢ 
                               
                                   
                               
                             
                           
                           ⁢ 
                           
                             ( 
                             i 
                             ) 
                           
                         
                       
                     
                   
                   I 
                 
                 . 
               
             
           
         
       
     
     
       6. The method of  claim 1 , wherein:
 the luminance component image corresponds to an input picture encoded based on a hierarchical structure, in which the input picture is hierarchically split from a plurality of maximum coding units, according to information about a maximum size of a coding unit, into coding units of coded depths according to depths; 
 a coding unit of a current depth is one of rectangular data units split from a coding unit of an upper depth; 
 the coding unit of the current depth is split into coding units of a lower depth, independently from neighboring coding units; and 
 coding units of the hierarchical structure comprise encoded coding units among the coding units split from a maximum coding unit. 
 
     
     
       7. An apparatus for predicting a chrominance component image by using a luminance component image, the apparatus comprising:
 a sampler configured to down-sample, by a processor, inner pixels of a luminance prediction unit and adjacent pixels of the luminance prediction unit, the luminance prediction unit having has been pre-encoded and pre-restored, considering a size of a luminance prediction unit and a size of a chrominance prediction unit corresponding to each other according to a color format; 
 a parameter obtainer configured to obtain, by a processor, parameters indicating a correlation between the luminance prediction unit and a predicted chrominance prediction unit based on the down-sampled adjacent pixels of the luminance prediction unit and restored adjacent pixels of the predicted chrominance prediction unit; and 
 a prediction performer configured to obtain, by a processor, from the down-sampled luminance prediction unit by using the obtained parameters, a prediction value of a chrominance prediction unit corresponding to the down-sampled luminance prediction unit, 
 wherein the sampler selects a first location of a luminance pixel to be selected from a predetermined luminance pixel group or a first filtering method of luminance pixels, for down-sampling with respect to the inner pixels of the luminance prediction unit, selects a second location of a luminance pixel to be selected from the predetermined luminance pixel group or a second filtering method of luminance pixels, for down-sampling with respect to adjacent upper pixels of the luminance prediction unit, and selects a third location of a luminance pixel to be selected from the predetermined luminance pixel group or a third filtering method of luminance pixels, for down-sampling with respect to adjacent left pixels of the luminance prediction unit, 
 wherein the sampler independently selects the first location and first filtering method used for down-sampling with respect to the inner pixels of the luminance prediction unit, the second location and second filtering method used for down-sampling with respect to the adjacent upper pixels of the luminance prediction unit, and the third location and third filtering method used for down-sampling with respect to the adjacent left pixels of the luminance prediction unit, 
 wherein, when the size of the luminance prediction unit is 2nS×2nS where nS is a positive integer: 
 the chrominance prediction unit corresponding to the luminance prediction unit is nS×nS, an inner pixel of the luminance prediction unit is RecL(x,y) where x and y are each from 0 to 2nS-1, an upper pixel of the luminance prediction unit is RecL(x,−1) where x is from 0 to 2nS-1, a left pixel of the luminance prediction unit is RecL(-1,y) where y is from 0 to 2nS-1, and a down-sampled luminance pixel corresponding to a chrominance pixel at (x,y) is RecL′(x,y); and 
 the sampler obtains a down-sampled luminance pixel RecL(x,−1) corresponding to an adjacent upper pixel Pc(x,−1) of the chrominance prediction unit according to equation, RecL′(x,−1)=(RecL(2x−1,−1)+2*RecL(2x,−1)+RecL(2x+1,−1)+2)>>2 or RecL′(x,−1)=RecL(2x, −1), with respect to adjacent top pixels of the luminance prediction unit, obtains a down-sampled luminance pixel RecL′(−1,y) corresponding to an adjacent left pixel Pc(−1,y) of the chrominance prediction unit according to equation, RecL′(−1,y)=(RecL(−1,2y)+RecL(−1,2y+1))>>1, with respect to adjacent left pixels of the luminance prediction unit, and obtains a down-sampled luminance pixel RecL′(x,y) corresponding to inner pixels Pc(x,y) of the chrominance prediction unit according to equation, RecL′(x,y)=(RecL(2x,2y)+RecL(2x,2y+1))>>1, with respect to the inner pixels of the luminance prediction unit. 
 
     
     
       8. A non-transitory computer readable recording medium having recorded thereon a computer program for executing the video encoding method of  claim 1 .

Cited by (0)

No later patents cite this yet.

References (0)

No backward citations on record.